Analysis of algorithms
In an interesting passage from the book The Shallows: What the Internet Is Doing to Our Brains by Nicholas George Carr develops the theme of freedom. In some way, Carr explains, these types of algorithms condition our freedom of choice, especially when they are explicitly developed to generate quasi-addictive behaviors. Are we still free? Or are we being shaped by the algorithm?
There is a lot of talk about how algorithms detect your digital behavior and preferences to then suggest content adapted to you and maintain online interaction. But little is said about the extent to which these decisions in the work of an algorithm are ethical. And, above all, who can assure us that this is the case?
The prestigious technosociologist Zeynep Tufeckci explained in a TED talk how complex algorithms are increasingly used to make consistent decisions about us and infer a large number of conclusions from the data about us on the Internet. Although this has a positive impact in certain scenarios, it can also end up being unfair in others. For this expert, the problem is that many of these decisions are subjective and do not have a single correct answer: who should be hired, fired or promoted; what news should be shown to whom; what updates from your friends should be shown in your account; which convict should be released on parole.
All of those questions are too delicate and leaving control to an algorithm could be counterproductive. One of the solutions would be precisely to be able to audit these algorithms and follow the lines of the European Union in terms of data protection laws. That would create a “right to explanation” that would allow consumers to know why an algorithm has made that decision that affects them and not another.
Given the growing prominence of AI, it is vital to ensure that algorithms are reasonable and representative of an organization’s values, as well as that they are aligned with current laws. Thus, the figure of the Algorithm Bias Auditor emerges with the aim of making this technology more explainable, transparent and controllable.